[XeTeX] No 'Number Case' in 'Hoefler Text Regular'?

[Jonathan Kew]

On 8 Jul 2005, at 10:42 am, Berteun Damman wrote:

> On 7/8/05, Will Robertson <will at guerilla.net.au> wrote:
>
>> When they were fixing some (but not all) of the bugs in Hoefler Text
>> for Tiger, they removed the font tables to access uppercase numerals
>> from the upright version of the font.
>>
>> I've reported the bug to them, but I imagine that it's not too high
>> on their priority list...
>>
>
> I've reported the bug too. But I thought, perhaps I can fix it myself,
> using the font tools of Apple (http://developer.apple.com/fonts/).
> But, that turned out to be a bit more involved than I thought.
> Apparently Apple switched to the 'morx' format for ligature
> substitution and such, in stead of mort, and Hoefler still uses
> 'mort'. Also, the utils do not expect this and they report that a font
> cannot have a feat table without a morx table. Sigh. So, I tried to
> dump the mort table to a MIF file using a tool, and deleting the feat
> and mort table, and building a new morx table. This fails horribly.
> Even when I only dump the mort table and use that one to reconstruct
> the morx table it does not give the original features.

If you really want to dig into the tables, there are Perl modules in  
the Font::TTF package (see CPAN) which can be used to disassemble and  
examine the 'mort' and other tables. But this is definitely  
programmer-level stuff, not end-user tools.

> But, I also discovered that the oldstyle rules *are* there and they
> *are* being dumped by ftxanalyser. It's only the 'feat' table that's
> not right. (But rebuilding it fails.) I assume this behaviour is a bug
> too and shall report it.

If I remember right, ftxanalyzer doesn't decompile the existing morx  
table, it writes suggested (default) MIF rules based on analysis of  
the glyph inventory in the font. So that may not be a valid test of  
whether the rules are actually present.

And in version 2.0, at least, of the Apple tools, the 'feat' table is  
built incorrectly, so that although the morx rules are present, the  
proper features and names are not displayed. I haven't yet tested  
whether this is fixed in the new release of the tools; I've been in  
the habit of post-processing the feat table to correct the name  
indexes, after compiling the MIF.

JK